How it works
What we read, what the three answers mean, and how we keep them honest.
The three answers
Every lookup ends in one of three. Which one you get changes what you should do next, so we never blur them together.
Found
We read the public profile. You get what it showed at that time: the name, the numbers and recent posts, with the source and original read time.
New profile read: 1 credit · cached answers are free
Not found
The site itself told us nobody uses that name. That is a real answer, not a failure, so it costs the same as a hit.
For a profile: 1 credit · cached for 1 hour
We could not check
We were blocked, the page timed out, or it would not load. We do not know, so we do not say. You get the credit back.
free · never kept
Why the third one exists
A profile that does not exist and one we could not read are different facts. Treating both as "not found" can make you draw the wrong conclusion about a person. That is why we keep the third answer separate.
What we read
We read public data ourselves and buy some answers from data providers, including posts and hashtag results. We do not ask for or use your social-network passwords or session cookies. Providers control their own collection methods. This is what each network can return.
instagram
Public profiles and their recent posts, with exact likes, comments and dates. If Instagram says the account is gone, we believe it.
tiktok
Public profiles with their latest videos: bio, verification and exact counts. A username nobody holds is a real “not found”.
youtube
Channels through the official API, then RSS, then the page. If a channel hides its subscriber count we return nothing, never zero.
linkedin
Company pages and personal /in/ pages. From a person we read the name, the headline and the exact follower count. A company page that does not exist is a real “not found”, from LinkedIn's own 404. A personal handle nobody holds comes back as “we could not check”, because LinkedIn answers a missing page and a blocked read with the same code, 999.
x
Public profiles from the page itself: bio, exact counts, verification and newest posts. A 404 is X's own answer.
facebook
Public pages and people: name, intro and the exact follower count. Facebook shows a page that doesn't exist and an 18+ page the same way, so we ask a second source. When neither can tell, we say we could not check instead of guessing.
github
Public users, organizations and repositories through GitHub's official API: bio, company, location, website, exact follower and repository counts, and the newest public repositories with stars, forks and topics. A login that does not exist answers 404, which is a real “not found”.
How we keep it honest
Three things that would be easy to skip, and that you would not be able to see us skipping.
A username search gets some wrong
A site can change and make its lookup rule wrong. Each night we test a batch of sites with a known registered name, starting with those checked longest ago. We try that name and one that cannot exist. Two failed test runs in a row exclude a site from searches by default.
Answers get old, and we say when
A profile is reused for 24 hours before we ask the source again. If a new read is refused, we may return the last good stored result at no charge, with no age limit. It keeps the original read time, not the time you requested it. We keep the data we read and its versions, not the original page or media files. Check the date before using an answer.
“We could not check” is never shown as “not found”
The API, the console and all three languages keep these answers separate. Each has its own label and color, so an unreadable profile is not presented as one that does not exist.
Try it on a handle you already know.
The fastest way to judge this is to look up an account you can check yourself.